|
View previous topic :: View next topic
|
| Author |
Message |
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Some folks of my acquaintenance royally stepped on their hand over the weekend. Their intent was to copy nearly all of their data from tape silos to Virtual Tape.
Many of the datasets were multi-volume files. When the copy jcl was generated, dataset names were pulled from a catalog list and put into individual steps. One volser was put into the input DD vol=ser.
The question is that if a dataset is catalogd with multiple tape volsers and then a job is run that reads the dataset, but specifies only 1 volser, will all of the cataloged volsers be used (up to eof) or only the one specified in the jcl?
I've looked in the JCL manuals and haven't found anything definitive and decided that asking here would be more productive than only spending time in the books (i'll continue looking - they're good people).
Thanks,
d |
|
| Back to top |
|
 |
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
I've done it on MVS. Just used the one tape, that specified on the VOL=SER= and got normal end, zero RC. I don't remember using the first volume, though, it was the last one I used a couple of times. IEBGENER I think it was I used. This was with UCCs tape management.
Aren't the JES (or whatever) messages going to give them the bad/good news? |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
No clarity, but I'm tending towards all the data being read correctly. You wanna nail me to an answer? I'd go with that. You can specify VOLUME=(,,vol-no) to start at a different volume than the first and continue normally, and I think that is equivalent to VOL=SER=vol-no. So, should be OK to specify the first volume as the start volume.
I agree, it is difficult to be definitive from the manuals I've looked at, but nothing really pointing to bad news. Everything kind of nearly says it, but doesn't quite make it.
Definitive, is the messages. |
|
| Back to top |
|
|
gcicchet
Senior Member
Joined: 28 Jul 2006
Posts: 1702
Location: Australia
|
|
|
|
Hi Dick,
the bad news is if you specify the vol ser in the JCL then only that volume will be copied.
For curiosity, what utility was used for copying the tapes ?
Our storage people have gone thru the same exercise using a third party product.
Everything is transparent, expriry dates creation dates, catalog entries are all taken care of.
The output JCL should show which volumes were mounted.
Also a quick copy of the original dataset and of the copied dataset can give a count of the records.
Gerry |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Thanks Bill 
Hi Gerry,
| Quote: |
| the bad news is if you specify the vol ser in the JCL then only that volume will be copied. |
Yup, that is what i thought but wasn't able to find it in the JCL docs. . . 
| Quote: |
Our storage people have gone thru the same exercise using a third party product. Everything is transparent, expriry dates creation dates, catalog entries are all taken care of.
|
The times i've been involved previously, it was rather automagic. . .
| Quote: |
| The output JCL should show which volumes were mounted. |
If only it was available. . .
This is the first time i've seen JCL submitted to copy a multi-volume datasdet that was cataloged and yet the first volser was specified in the jcl. I've suggested they get the catalog back to where it was before the weekend (for these series of datasets) and i'll help move it forward.
The few i was told about were done with IEBGENER. I'll probably use Syncsort. Once upon a time (more than 10 years ago), i seem to recall using FAVER.
d |
|
| Back to top |
|
|
gcicchet
Senior Member
Joined: 28 Jul 2006
Posts: 1702
Location: Australia
|
|
|
|
Hi Dick,
the SYSLOG should show which tapes were mounted but it will only confirm what I have already mentioned earlier. 
Gerry |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
Gerry,
Just out of interest, that would mean that "VOLUME=(,,vol-no)" is not equivalent to "VOL=SER=vol-no". Any idea why not?
This is specified with the catalogued dataset name. The manual definitely says the first will work with other than the first volume (I guess they didn't imagine they'd have to be clear about the first volume).
Dick,
If SYSLOG also missing/not easily available, contact the OPS and ask them how tired they were are the weekend. Suggest to them that it might have to be done again. Judge from response how many mounts they did :-) Or ask them if jobs asked for more than one tape. Or is it done by robots? |
|
| Back to top |
|
|
gcicchet
Senior Member
Joined: 28 Jul 2006
Posts: 1702
Location: Australia
|
|
|
|
Hi Bill,
vol=(,,seqno) - it's the sequence number not the vol ser
| Quote: |
vol seq #
provides the volume number of a multi-volume data set at
which processing is to begin for this DD statement - it may
be a decimal number in the range 1 through 255, inclusive -
this parameter is ignored for new data sets
|
Gerry |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
| Thanks Gerry. More haste, less speed. More actual reading. |
|
| Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
Been thinking about this whilst travelling around this morning. Must be several scrillion times I wanted something specific from one volume of a multi-volume tape dataset. VOL=SER=. Then there was the time I wanted to read the file but starting at the last volume. Used VOL=SER= and it "worked". Only now (this was sometime in the early '80s, New Year's Day, a Saturday, quarter to midnight, no ops shift planned for the Sunday and the Monday a public holiday. Can't get to the actual year now, as Windows doesn't go back that far) I realise it only "worked" by accident. But, then and all the time since I have segregated the two "different" situations in my memory. Dumb, huh?
Must be some fancy name for the syndrome. Or just Dumb will do.
Dick, did you mention "testing" or "post-implementation" to the offending parties? Also, when we warned the OPS that we would have big tape usage, and we didn't, they always let us know. It was a big, big clue that something was wrong. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hello,
| Quote: |
| did you mention "testing" or "post-implementation" to the offending parties? |
Someone in a positon to generate a crisis did so nd now they are all paying the price. There was almost no planning and no verification steps to be done before the wholesale runs were done.
All of the 3490 data is in robots, so the operators don't have to do days of the "Cartridge Shuffle". . .
Just to add a bit more - when someone realized they had problems with multi-volume files, they decided to copy each of the multiple volumes to a new single-volume file. How awful. . . My suggestion this morning has been that they re-catalog the original volsers, create and test new copy jcl, and then move forward.
Fortunately, this exercise is less than 10,000 carts. An earlier exercise far away was to migrate about 2 million carts. . .
Let's see what new discoveries are made before the re-run. . .
'Preciate the inputs
d |
|
| Back to top |
|
|
|
|
